1. Field of the Invention
The present invention relates to an apparatus and a method for forming a conductive polymer bump on a substrate, such as a flip-chip type semiconductor die, a silicon wafer, a printed circuit board, or other substrate (hereinafter referred to generally as a "substrate"). More particularly, the present invention relates to forming a substrate having a palladium metal layer over each contact point of the substrate and forming a flexible conductive polymer bump on each contact point. The present invention also relates to assemblies and methods of connecting one or more of these substrates together or to another substrate.
2. State of the Art
A flip chip is a semiconductor chip or die that has bumped terminations spaced around an active surface of the die and is intended for face-to-face attachment to a substrate or another semiconductor die. The bumped terminations of the flip chips are usually a "Ball Grid Array" ("BGA") configuration wherein an array of minute solder balls is disposed on an attachment surface of a semiconductor die, or a "Slightly Larger than Integrated Circuit Carrier" ("SLICC") configuration wherein an array of minute solder balls is disposed on an attachment surface of a semiconductor die similar to a BGA, but having a smaller solder ball pitch and diameter than a BGA.
The attachment of a flip chip to a substrate or another semiconductor involves aligning the solder balls on the flip chip with a plurality of contact points (configured to be a mirror image of the solder ball arrangement on the flip chip) on a facing surface of the substrate. A plurality of solder balls may also be formed on the facing surface of the substrate at the contact points. A quantity of liquid flux is often applied to the face of the chip and/or substrate, and the chip and substrate are subjected to elevated temperature to effect reflowing or soldering of the solder balls on the chip and/or corresponding solder balls on the substrate. This connection technology is also referred to as "flip chip attachment" or "C4-Controlled Collapse Chip Connection."
High performance microelectronic devices generally comprise a number of flip chips, attached to a substrate or printed circuit board ("PCB") for electrical interconnection to other microelectronic devices. For example, a very large scale integration ("VLSI") chip may be electrically connected to a substrate, printed circuit board, or other next level packaging substrate.
Flip chip attachment requires the formation of contact terminals on flip chip contact sites, each consisting of a metal pad with a solder ball disposed thereon. Flip chip attachment also requires the formation of solder joinable sites ("bond pads") on the metal conductors of the substrate or PCB which are a mirror-image of the solder ball arrangement on the flip chip. The bond pads on the substrate are usually surrounded by non-solderable barriers so that when the solder of the bond pads and of the chip contact sites melts and merges ("reflow"), the surface tension holds the semiconductor chip by solder columns, as if suspended above the substrate. After cooling, the chip is essentially welded face-down by these very small, closely spaced solder column interconnections.
It is also known in the art that conductive polymers or resins can be utilized in lieu of solder balls. U.S. Pat. No. 5,258,577 issued Nov. 2, 1993 to Clements relates to a substrate and a semiconductor die with a discontinuous passivation layer. The discontinuities result in vias between the contact points of the substrate and the semiconductor die. A resin with spaced conductive metal particles suspended therein is disposed within the vias to achieve electrical contact between the substrate and the semiconductor die. U.S. Pat. No. 5,468,681 issued Nov. 21, 1995 to Pasch relates to interconnecting conductive substrates using an interposer having conductive plastic filled vias. U.S. Pat. No. 5,478,007 issued Dec. 26, 1995 to Marrs relates to using conductive epoxy as a bond pad structure on a substrate for receiving a coined ball bond on a die to achieve electrical communication between the die and the substrate.
Such flip chip and substrate attachments (collectively "electronic packages") are generally comprised of dissimilar materials that expand at different rates on heating. The most severe stress is due to the inherently large thermal coefficient of expansion ("TCE") mismatch between the plastic and the metal. These electronic packages are subject to two types of heat exposures: process cycles, which are often high in temperature but few in number; and operation cycles, which are numerous but less extreme. If either the flip chip(s) or substrate(s) are unable to repeatedly bear their share of the system thermal mismatch, the electronic package will fracture, which destroys the functionality of the electronic package.
As an electronic package dissipates heat to its surroundings during operation, or as the ambient system temperature changes, differential thermal expansions cause stresses to be generated in the interconnection structures (e.g., solder ball bonds) between the semiconductor die and the substrate. These stresses produce instantaneous elastic and, most often, plastic strain, as well as time-dependent (plastic and anelastic) strains in the joint, especially within its weakest segment. Thus, the TCE mismatch between chip and substrate will cause a shear displacement to be applied on each terminal which can fracture the solder connection.
The problem with TCE mismatch becomes evident during the process of burn-in. Burn-in is the process of electrically stressing a device, usually at an elevated temperature and voltage environment for an adequate period of time to cause failure of marginal devices. When a chip, such as a flip chip, breaks free from the substrate due to TCE mismatch, defective bonds, or the like, the chip must be reattached and the burn-in process reinitiated. This requires considerable time and effort which results in increased production costs. Alternately, if the chip has been underfilled and subsequently breaks free during burn-in, the chip in not reworkable and must be discarded.
The problems with TCE mismatch are also applicable to connections made with conductive polymers or resins, because after curing the polymers or resins become substantially rigid. The rigid connections are equally susceptible to breakage due to TCE mismatch.
FIGS. 1a-1e show a contemporary, prior art method of forming a conductive bump arrangement on a substrate. First, as shown in FIG. 1a, a passivation film 102, such as at least one layer of SiO.sub.2 film, Si.sub.3 N.sub.4 film, or the like, is formed over a face surface 104 of a semiconductor wafer 100 which has a conductive electrode 106, usually an aluminum electrode. The passivation film 102 is selectively etched to expose the conductive electrode 106. FIG. 1b shows a metal layer 108 applied over a face surface 110 of the passivation film 102 by deposition or sputtering. A second layer of etch resist film 112 is applied to a face surface 114 of the metal layer 108. The second etch resist film 112 is masked, exposed, and stripped to expose a portion of the metal layer 108 corresponding to the conductive electrode 106, as shown in FIG. 1c. A solder bump 116 (generally an alloy of lead and tin) is then formed on the exposed portion of the metal layer 108, as shown in FIG. 1d, by any known industry technique, such as stenciling, screen printing, electroplating, electrolysis, or the like. The second etch resist film 112 is removed and the metal layer 108 is removed using the solder bump 116 as a mask to form the structure shown in FIG. 1e. This conventional bump formation method has drawbacks. The most obvious being the large number of process steps required which results in high manufacturing costs.
U.S. Pat. No. 4,970,571 issued Nov. 13, 1990 to Yamakawa et al. (the '571 patent) relates to a bump formation method which addresses the problems associated with conventional processing methods by using electroless plating of palladium on the conductive electrodes. Electroless plating is a metal deposition process, usually in an aqueous medium, occurring through an exchange reaction between metal complexes in solution and the particular metal to be coated which does not require externally applied electric current. The process of electroless plating of palladium generally comprises dipping the semiconductor element with the exposed conductive electrodes into a palladium solution wherein the palladium selectively bonds or plates on the conductive electrodes. The electroless plating process is a non-vacuum, high volume, high throughput process which can be precisely controlled and uses reliable equipment. The entire fabrication can be performed in a less costly cleanroom environment which reduces processing time and cost. The '571 patent teaches forming an electroless palladium plated conductive electrode followed by the formation of a metal bump on the palladium plated conductive electrode.
The benefits of using palladium in integrated circuits are discussed in U.S. Pat. No. 4,182,781 issued Jan. 8, 1980, to Hooper et al. (the '781 patent). The '781 patent teaches that palladium forms a unique barrier metal in bump metallization systems which increases yield and reliability of integrated circuit devices designed for flip-chip attachment. It is also disclosed that palladium is compatible with aluminum and has a thermal coefficient of expansion that is sufficiently close to aluminum that no significant stress problems result. However, the '781 patent does not teach using an electroless plating process to coat the conductive electrode with palladium. However, electroless plating is used to form the copper or nickel bump on the palladium coated conductive electrode.
It would be advantageous to develop a more efficient technique for forming conductive bumps on a flip chip which eliminates some of the steps required by present industry standard techniques while also abating the effects of TCE mismatch using commercially-available, widely-practiced semiconductor device fabrication techniques.